Company being whatever pharmaceutical manufacturer was funding the study. Reasons for an incomplete? Sometimes members of the trial used recreational drugs or simply dropped out.
Those keeping the books were coereced, on occasion, to doctor the stats, including the minimum number of eligibile patients to ascertain a valid result.
Two patients on a drug screen tested positive for methamphetamine compramising eligibility? Make sure not to scan that in, keep paper copies for reference..
Into the trash they went.
These things do tend to get exposed. There is also the issue of "p hacking" to try and make it seem as if a result has statistical significance when has not met the threshold.
Thankfully this all tends to get exposed. This is why the system works so well. The lies in the stakes get exposed. They never get exposed by a random guy on the internet, but research
